Groningen AI Factory to Get Less Computing Power Due to Pricier Nvidia Chips

19 August 2026 · 06:00 · Claude (Anthropic) · claude-sonnet-5

The long-awaited Groningen AI factory will be less powerful than planned, as chips from Nvidia and AMD have become significantly more expensive due to global demand for AI hardware. The fixed budget of 64 million euros for the supercomputer therefore buys less computing power than originally expected.

The Groningen AI factory was supposed to give the Netherlands its own powerful supercomputer for artificial intelligence, but the project is being forced to scale back its ambitions. The cause is simple but stubborn: AI chips from market leader Nvidia and rival AMD have become significantly more expensive in recent months due to explosive global demand for AI computing power. Because the available budget is fixed, this inevitably means fewer chips and therefore less computing power than initially promised.

What exactly is the Groningen AI factory?

The AI factory is a major infrastructure project in the northern Netherlands, with a supercomputer as its beating heart. A total of 201 million euros has been set aside for its construction and setup: 60 million euros comes from the Nij Begun earthquake recovery program, 70 million euros from the Dutch government, and 71 million euros from European subsidies. Locations such as the Eemshaven and Zernike Campus have been proposed because of their reliable power supply, fiber-optic connections, and cooling options, all crucial for a data center that needs to run heavy AI training workloads.

The goal is ambitious: the factory should give Dutch startups, research institutions, and companies access to computing power that is currently concentrated mainly with American tech giants and their cloud platforms. Supporters point to thousands of potential new jobs and hundreds of startups that could benefit from the facility.

Why are the chips becoming so much more expensive?

A fixed budget of 64 million euros is available for purchasing the supercomputer. That amount has been locked in for some time, but the prices of the required components have since risen sharply. Major tech companies around the world are currently pouring enormous sums into their own AI infrastructure, causing demand for advanced GPUs, memory chips, and storage to far outstrip available production capacity. As a result, some components have become two to several times more expensive in a short period.

Nvidia as the deciding factor

With its powerful GPUs, Nvidia has been the dominant supplier for AI training for years, and public projects like the Groningen factory are feeling that dominance too. The supercomputer was originally meant to include around 1,800 advanced GPUs to train large language models. The exact number of chips that will ultimately be purchased depends on the procurement process that is still underway, but due to the price increases, that number is expected to fall short of the original plan.

Consequences for computing power and planning

Fewer chips for the same budget directly translates into less computing power for the factory. That is a problem for a project specifically intended to make Dutch organizations less dependent on foreign cloud infrastructure. A less powerful supercomputer could mean that larger AI models are trained more slowly, or that fewer users can access the facility at the same time. The parties involved stress that the project is moving forward, but discussions about what is and isn't feasible within the current budget are ongoing.

What does this mean for Dutch AI ambitions?

The story out of Groningen is emblematic of a broader problem: governments and public institutions that want to compete in AI are running into a market where Nvidia, AMD, and other chipmakers set prices based on demand from commercial tech giants. The Netherlands is not the only country facing this; other European AI gigafactories are dealing with similar cost increases. Anyone wanting to put the development of this kind of technology into perspective can read more about the history of artificial intelligence and how computing power has always been a decisive factor in AI's progress.

Conclusion

The Groningen AI factory remains an important symbol of the Dutch ambition to invest independently in AI infrastructure, but the case shows how vulnerable such plans are to a volatile chip market. As long as global demand for Nvidia and AMD chips remains sky-high, public projects with fixed budgets will also have to make compromises. Curious how AI infrastructure will continue to evolve and which AI applications it will enable? Follow more AI news and dive into our knowledge base for background on the latest developments.

Volkskrant

Source: Volkskrant